Using the Asphalt Pavement Analyzer to Assess Rutting Susceptibility of HMA Designed for High Tire Pressure Aircraft

Transcription

1 Using the Asphalt Pavement Analyzer to Assess Rutting Susceptibility of HMA Designed for High Tire Pressure Aircraft Submission date: November, Authors: John F. Rushing (corresponding author) U.S. Army Engineer Research and Development Center CEERD-GM-A 0 Halls Ferry Road Vicksburg, MS - Phone: (0) - Fax: (0) John.F.Rushing@usace.army.mil Dallas N. Little, Ph.D. Texas A&M University CE/TTI 0 TAMU College Station, TX - Phone: () - Fax: () - d-little@tamu.edu Navneet Garg, Ph.D. Federal Aviation Administration Airport Technology R&D Branch William J. Hughes Technical Center Atlantic City Intl. Airport, NJ 00 Phone: (0) - Fax: (0) - Navneet.Garg@faa.gov 0 Word Count Total:, Text:, Tables and Figures (*0):,0 TRB 0 Annual Meeting

2 Rushing, Little, and Garg ABSTRACT Hot Mix Asphalt (HMA) laboratory mix design is intended to determine the proportion of aggregate and binder that, when mixed and compacted under a specified effort, will withstand anticipated loading conditions. Current mix design procedures using the Superpave Gyratory Compactor (SGC) rely on engineering properties and volumetrics of the compacted mixture to assure reliable performance; however, a definitive performance test does not exist. The Asphalt Pavement Analyzer (APA) is evaluated as a tool to assess HMA mixtures designed to perform under high tire pressure aircraft following Federal Aviation Administration (FAA) specifications. The APA used in this study was specially designed to test simulated high tire pressures of 0 psi, common for aircraft. Thirty-three HMA mixtures are included in the study. Each was designed using the SGC according to preliminary criteria being developed by the FAA. The study included some mixtures that contained excessive percentages of natural sand and that do not meet FAA criteria. These mixtures are included to provide relative performance for mixtures expected to exhibit premature rutting. Asphalt Pavement Analyzer testing using the high tire pressure APA resulted in rapid failure of HMA specimens compared to traditional APA testing at lower pressures. Data were analyzed with the focus of providing acceptance recommendations for mixtures to support high tire pressures. A preliminary mm rut depth criterion after,000 load cycles is recommended. 0 0 TRB 0 Annual Meeting

3 Rushing, Little, and Garg 0 0 INTRODUCTION Rutting is a primary load-related distress in airport pavements subjected to high tire pressure aircraft loads. Hot Mix Asphalt (HMA) for airport pavements has historically been designed using materials and compaction requirements that can withstand these loading conditions. The Marshall design procedure has been used by the Federal Aviation Administration (FAA) for airport pavement. The design procedure includes meeting aggregate and binder requirements along with the laboratory mix design volumetric requirements to produce acceptable mixtures. Marshall stability and flow test values provide an empirical, threshold metric to help insure a stable mixture under traffic, but this empirical test cannot assess performance. Performance can be assessed by either a damage model that is based on sophisticated testing of engineering and material properties or tests that mimic the traffic loads encountered in the field. A new mix design procedure being implemented by the FAA allows the use of the Superpave Gyratory Compactor (SGC) in lieu of Marshall impact compaction to compact specimens during the mix design process (). This new procedure includes the same aggregate and binder requirements and the same volumetric requirements as those that are currently part of the Marshall protocol, but no performance test is included in the protocol. Studies are currently being conducted to identify a companion performance test for the new mix design procedures, but one has not yet been adopted. The Asphalt Pavement Analyzer (APA) has been successfully used by several agencies as a test to determine HMA mixture suitability and is one of the most widely accepted laboratory accelerated wheel trafficking devices available and as one that attempts to mimic the action of a moving and heavily loaded, high tire pressure wheel of the type that simulates airfield traffic. Tests that simulate or mimic traffic action are often used in lieu of traditional laboratory testing because the action of a moving wheel load in which there is a rotation of principal stresses and a transition from compression to extension is difficult to recreate with even sophisticated triaxial laboratory tests (). For this reason, empirical tests such as the APA offer an alternative approach, which is favored by many. The APA was first manufactured in (). It places a loaded wheel on a pressured linear tube, centered on top of the specimen. The wheel is tracked in a forward and backward linear motion across the sample(s), resulting in plastic deformation, or rutting. The APA can be used to test either cylindrical or beam specimens. Several studies have attempted to correlate APA laboratory rutting with field rutting. Williams and Prowell () found that the APA test results correlated well with field results, as did a number of state departments of transportation (-). However, as the WesTrack Forensic Team Study demonstrated, a direct relationship between lab rutting and field rutting does not exist. For example the WesTrack study showed that a laboratory rut depth of mm after,000 cycles represented a field rut depth of. mm (). Choubane, Page, and Musselman () reported appropriate ranking of mixes according to field performance. However, their study indicated APA test variability was statistically significant from test to test and among locations during a test. Recommendations from the study cautioned on using the APA as TRB 0 Annual Meeting

4 Rushing, Little, and Garg 0 0 a pass or fail criterion based on the three mixes tested. AASHTO TP -0 indicated precision and bias statements have yet to be determined for APA testing. OBJECTIVE The objective of this study was to evaluate the ability of the APA to serve as a performance test to assess the potential for rutting under high tire pressure aircraft during the design of hot asphalt mixtures. HMA mixtures carefully designed to exhibit a wide range of rutting potential were included in the study. The ability of the APA to differentiate among these mixtures and its applicability as a mix design and/or quality assurance test were investigated. The actual rutting susceptibility of the mixtures tested under field conditions was not evaluated; however, mix variables known to contribute to rutting were controlled in order to provide relative performance correlations. TEST PROCEDURE The APA used in this study was specifically designed to simulate high tire pressures associated with aircraft. Two such high pressure APAs are known to exist in July 0: one at the FAA William J. Hughes Technical Center in Atlantic City, New Jersey, and one at the Air Force Research Laboratory at Tyndall Air Force Base, Florida. Testing in this study was conducted at the FAA test facility. Mix design and specimen preparation were accomplished at the U.S. Army Engineer Research and Development Center (ERDC), Waterways Experiment Station in Vicksburg, Mississippi. An APA tube or hose pressure of 0 psi under a wheel load of 0 lb was used for testing. The test temperature was o C as was the high temperature PG grade for the binder used in all mixtures. Cylindrical asphalt concrete specimens with a target air void content of.% were prepared and tested. The air void content was selected as the midpoint of the allowable range in the FAA mix design procedure. Two replicate specimens were tested for each mix. The APA reports the average rut depth of the two specimens. Thirty-three mixtures were included in the study. The mixtures used one neat binder and three aggregate types. Some mixtures included natural sand. Cyclic loads were applied by the APA at a rate of one cycle per second. The terminal rut depth of the specimens was set at mm after,000 cycles; however, the test was terminated when the mm rut depth was achieved if this occurred before,000 cycles. Once one of the two specimens reached terminal rut depth, the test was stopped. However, since the APA reports the average rut depth for the two specimens, some average rut depths were less than mm. MATERIALS The asphalt binder used in this study was obtained from Ergon Asphalt and Emulsions, Inc. Tests by the distributor indicated the binder graded as a PG - and had a specific gravity of.0. Supplierrecommended mixing and compaction temperatures for the binder were o F ( o C) and 0 o F ( o C), respectively, based on binder viscosity. Aggregates used in this study consisted of material stockpiles at the ERDC collected for a previous study which evaluated the use of the SGC for airport mix design (). These aggregates included limestone, granite, and chert gravel. The limestone aggregate was from a Vulcan Materials quarry in TRB 0 Annual Meeting

5 Rushing, Little, and Garg 0 0 Calera, Alabama. The granite aggregate was from Granite Mountain Quarries in Little Rock, Arkansas. The chert gravel aggregate was from Green Brothers Gravel Company in Copiah County, Mississippi. Additionally, some mixtures were blended with selected percentages of natural sand, which was locally purchased from Mississippi Materials Corporation in Vicksburg, Mississippi. Each aggregate type was comprised of multiple stockpiles that were blended to meet the target gradations. Selected gradations were within the allowable range of size fractions incorporated in FAA specifications (). The gradations referred to in this paper are designated as fine and coarse. Fine gradations are those near the upper limits of the gradation band. Coarse gradations are those near the lower limits of the gradation band. Some aggregate blends included or 0 percent natural sand. These gradations are characterized by a hump in the grain size distribution near the 0 to 0 sieve sizes. The percentages of aggregate with at least two fractured faces were 0, 0, and percent for the limestone, granite, and chert gravel, respectively. The maximum percentages of flat and elongated aggregates were.,.0, and 0. percent for the limestone, granite, and chert gravel, respectively. The fine aggregate angularity for the limestone, granite, chert gravel, and mortar sand aggregates was determined by Method A of ASTM C. The limestone, granite, and chert gravel aggregates had a fine aggregate angularity of,, and percent, respectively. The fine aggregate angularity of the mortar sand was 0 percent. This value is characteristic of rounded aggregate particles and is typical for natural sands (). Additional testing of the aggregates was performed using the Aggregate Imaging System (AIMS). AIMS determines shape characteristics of aggregates through image processing and analysis techniques (). AIMS is a computer automated system that includes a lighting table where aggregates are placed in order to measure their physical characteristics (shape, angularity and texture). It is equipped with an autofocus microscope and a digital camera, and is capable of analyzing the characteristics of aggregates sizes retained on the #0 sieve (0. mm sieve) up to aggregates retained on the inch sieve (. mm). Texture is measured by analyzing gray scale images captured at the aggregate surface using the wavelet analysis method. The surface irregularities manifest themselves as variations in gray-level intensities that range from 0 to. Large variations in gray-level intensity mean a rough surface texture, whereas a smaller variation in gray-level intensity means a smooth particle. The wavelet transform analyzes the image as a two dimensional signal of gray scale intensities, and it gives a higher texture index for particles with rougher surfaces. Angularity is measured using the gradient analysis method which basically quantifies the change in angles along the circumference of a particle. A higher change in angle means a more angular particle. Masad et al. () gives detailed background information about AIMS operations and analysis methods. Six size fractions for each of the three aggregate types were tested using AIMS. These fractions included aggregates retained on / in., / in., #, #, #, and #0 U.S. standard sieves during a washed sieve analysis. Table shows the average results for angularity and texture indices measured by AIMS. TRB 0 Annual Meeting

6 Rushing, Little, and Garg TABLE Aggregate data from AIMS analysis Sieve Angularity Texture Size Limestone Granite Chert Gravel Limestone Granite Chert Gravel / in / in. # 0 0 # 0 # 0 #0 Texture is only measured on coarse aggregate 0 MIX DESIGN A Pine Instruments Company model AFGCX gyratory compactor was used in the mix designs during this study to produce cylindrical asphalt concrete specimens with a diameter of 0 mm at a target height of mm. Compaction was performed using a ram pressure of psi (00 kpa) and an internal angle of gyration of. o ± 0.0 o. Asphalt mixtures were compacted to 0 gyrations at a rate of 0 revolutions per minute. Seventy gyrations is recommended for N design for HMA mixtures designed for high tire pressure aircraft (), but this is currently being evaluated by the FAA. Three different binder contents were used for the mix designs, in increments of 0. % binder. Three replicate specimens were compacted at each binder content. The mix designs were able to bracket the design binder content using only three trial percentages because of previous experience with these materials (). The air void content and voids in mineral aggregate (VMA) were determined according to Asphalt Institute MS-0 procedures (). The design or optimal binder content was selected as the binder content that resulted in a compacted specimen having.% air voids. Specimens produced for APA testing were compacted using the design binder content at a target height of mm. The mass of binder and aggregate was proportioned so that the compacted specimen height would be near the target value. Reducing the mass of the material in the mold was expected to have some influence on the volumetric properties. The smaller sample size, along with inherent specimen variability, resulted in some specimens having higher or lower air voids than the target value of.%. Table shows the aggregate types, maximum size, and relative gradation as well as the percentage of natural sand, the binder content, and the air void content for the APA test specimens. The mix designation is also listed and is used to identify these mixtures in this paper. TRB 0 Annual Meeting

7 Rushing, Little, and Garg TABLE APA test specimen characteristics Results from Superpave Mix Design at 0 Gyrations (.% Air Voids) Aggregate Type Maximum Aggregate Size Gradation Percentage of Mortar Sand Optimum Binder Content Air Voids Air Voids Mix Designation 0... / FGN Fine..0. / FGN /" Coarse 0... / FGN / CGN... / CGN Granite Fine 0... / CGN / FGN.. 0. / FGN /" Coarse / FGN / CGN... / CGN / CGN / FLS Fine...0 / FLS /" Coarse / FLS / CLS.0.0. / CLS Limestone Fine / CLS / FLS..0.0 / FLS /" Coarse / FLS / CLS..0.0 / CLS 0... / CLS / FGV /" Center..0. / FGV 0... / FGV 0 Chert Gravel /" Fine Coarse 0... / FGV..0. / FGV 0... / FGV / CGV... / CGV 0... / CGV 0 TRB 0 Annual Meeting

8 Rushing, Little, and Garg APA RESULTS General The APA records the average rut depth of the two specimens with each load cycle into a Microsoft Excel spreadsheet. The average rut depth was plotted versus the number of load cycles to produce a curve of accumulated rutting. Figure shows the number of load cycles versus APA rut depth for all mixtures. As previously stated, the test was ceased after one of the two specimens reached the terminal rut depth of mm. The values shown in Figure are the average rut depth of the two specimens. The mix designation is not included on this figure because of the large number of data sets. Data from Figure are extracted in subsequent figures with mix designations included for analysis. Rut Depth (mm) Load Cycles FIGURE APA results for all mixtures During APA testing, a more rapid rutting rate took place during the initial load cycles. After about mm of rut depth, specimen behavior was observed to be more closely linked to mixture characteristics as evidenced by variable rates of rut depth accumulation. Some mixtures had a slow rate of rut depth accumulation while others failed very quickly. The rate of rut depth accumulation appeared to become somewhat linear after approximately mm of rutting. The rutting behavior in the APA TRB 0 Annual Meeting

9 Rushing, Little, and Garg 0 followed the same general pattern as commonly observed in creep and repeated loading experiments with a primary and secondary flow. Tertiary flow was not observed during the experiments. Although mix designations are not shown in Figure, analysis of the data indicates general trends in mix variables that impact the rate of rut depth accumulation. These mix variables are investigated in the following paragraphs. An analysis was conducted to determine the effect of natural sand on the APA results. Figure shows the APA results for mixtures containing no natural sand. These mixtures, as expected, were among the best performers in the APA test. Incorporating natural sand promotes rutting in HMA (, ). For mixtures containing no natural sand, the accumulation of rut depth is related to the aggregate type. Mixtures containing crushed chert gravel aggregate rutted much more quickly than the other mixtures. The crushed gravel meets the FAA requirements for the mass percentage of aggregate particles having at least two fractured faces (0% for coarse aggregate). However, these aggregates also have low levels of angularity and relatively smooth texture. Interparticle friction, although not directly measured, is accepted to be lower for chert gravel than for quarried aggregate. In addition, chert gravel mixtures commonly have a higher VMA than quarried aggregate mixtures, resulting in more binder required to compact mixtures to equivalent air void content. Mixtures containing crushed limestone aggregate performed best in the APA test. In general, the crushed granite mixtures rutted more quickly than the crushed limestone mixtures. The crushed granite mixtures, on average, had a higher design binder content than the crushed limestone mixtures. The differences in design binder content or binder demand are likely to be influenced by factors such as aggregate shape, texture, and breakdown during compaction or mixture VMA. 0 TRB 0 Annual Meeting

10 Rushing, Little, and Garg / FGV / GV / FGN / CGN / CGV / FGN / FLS / GV / FGV Rut Depth (mm) / CGN / CLS / FLS / CLS / CGV / FLS / CLS / FLS / CLS / FGN / CGN / FGN / CGN Load Cycles FIGURE APA results for mixtures containing no natural sand Figure shows the APA results for mixtures containing % natural sand. These mixtures rutted more quickly than those containing no natural sand. Once again, mixtures produced with crushed gravel rutted more quickly than other mixtures. Similarly, mixtures produced with crushed limestone performed best by demonstrating the greatest resistance to rutting. TRB 0 Annual Meeting

11 Rushing, Little, and Garg / FGN / FGV / CGV / CLS / GV / CGN / FLS / GV Rut Depth (mm) / CLS / CGN / FGN / FLS / FGV / CGV / FLS / CLS / FLS / CLS / FGN / CGN / FGN / CGN Load Cycles FIGURE APA results for mixtures containing % natural sand Some mixtures contained 0% natural sand, a higher percentage of natural sand than is allowed by FAA specifications (maximum of %). The mixtures were included in the analysis because they were expected to rut quickly and could provide guidance for performance threshold levels within the specifications. All mixtures containing 0% natural sand failed quickly (less than,00 cycles) in the APA. In general, the mixture variable with greatest influence on APA test results was the percentage of natural sand. High percentages of natural sand caused premature failure in the APA. Additionally, aggregate type influenced the APA test results. Mixtures of chert gravel rutted more quickly than mixtures with granite or limestone aggregate. Statistical Considerations Analyses were performed to evaluate the impact of mixture variables on rut depth accumulation considering statistical significance. All analyses were performed using SigmaStat software at a % confidence level. The analysis of variance (ANOVA) procedure, including the Tukey test for all pairwise comparison, was used to evaluate data sets. The number of load cycles to reach an average rut depth of TRB 0 Annual Meeting

12 Rushing, Little, and Garg 0 0 mm was used as a metric to quantify mixture behavior in these analyses. A rut depth of mm was selected because it is near the maximum average rut depth reported in the APA samples. The statistical analyses considered the following factors: () the influence of aggregate type, i.e., limestone, chert gravel or granite; () the influence of shape and textural factors (angularity, surface texture and flat and elongated shape factors); and () the impact of the presence and amount of field or natural sand (uncrushed). First, the influence of aggregate type on APA results was investigated. Granite, limestone, and chert gravel mixtures containing no natural sand require statistically different numbers of load cycles to reach mm of rutting in the APA. The average number of load cycles required to reach mm rutting for these respective mixtures was,0; 000; and,0. It is important to note that the value of,000 cycles is the terminal test value and no data were extrapolated. Further analyses on the influence of aggregate type on APA rutting were performed on the AIMS data. The indices of shape and angularity were investigated and considered to be the aggregate characteristics most closely related to rutting in HMA. Aggregate angularity was measured for three sizes of coarse aggregate and three sizes of fine aggregate for each aggregate type using AIMS. Coarse and fine aggregate were compared independently using ANOVA. Angularity was not found to be statistically different among the three size fractions within a specific aggregate type. The difference between the coarse aggregate fraction of limestone and chert gravel aggregates was not statistically significant. For each coarse size fraction, the granite aggregate was more angular than limestone. Granite was statistically more angular than chert gravel aggregate on ½ in. and # sieve sizes, but not the / in. sieve. Similarly, no statistically significant difference in angularity was detected between limestone and chert gravel fine aggregate. All size fractions of fine granite aggregate were statistically more angular than all size fractions of limestone or chert gravel aggregate. Aggregate texture is only measured on coarse aggregate by AIMS. Statistical analyses of AIMS texture data indicate all sizes of chert gravel have a statistically lower texture index than any size fraction of limestone or granite aggregate. For all equal size fractions, granite aggregate has a higher texture than limestone aggregate. In summary, the AIMS data rank the aggregates the same by both angularity and texture, with granite having the highest indices and chert gravel having the lowest indices, although the angularity of limestone and chert gravel are very similar. Higher angularity and texture indices are expected to result in greater rutting resistance. On this basis, the degree of angularity and texture are consistent in terms of identifying chert gravel mixtures as the most rut susceptible, but inconsistent in predicting granite as the most rut resistant mixtures. APA results indicate aggregate texture may be a better indicator of rutting resistance than angularity. The lower resistance of the granite mixtures to APA rutting compared to limestone mixtures likely results from higher design binder contents. In fact, a Pearson s product moment correlation analysis considering the influence of aggregate angularity and texture, mixture binder content, and specimen air void content on APA rutting indicated aggregate texture and binder content were the only statistically influential variables. Higher rut resistance resulted from increased aggregate texture and lower binder content. TRB 0 Annual Meeting

13 Rushing, Little, and Garg 0 0 The addition of neither % nor 0% natural sand to the mixtures impacted the rank order of rutting sensitivity among the three aggregate types. For mixtures containing % natural sand, the mean number of load cycles resulting in mm rut depth was,;,0; and,0 for granite, limestone, and chert gravel mixtures, respectively. The trend of performance, according to the number of load cycles resulting in mm rutting, was the same for mixtures containing no natural sand as for mixtures containing % natural sand. Limestone mixtures performed the best while chert gravel mixtures performed the poorest. However, only limestone and chert gravel were statistically different from each other. For mixtures containing 0% natural sand, the numbers of load cycles to mm rut depth were 0, 0, and, respectively. Each of these mixtures failed rapidly compared to mixtures with a higher percentage of crushed aggregate. The effect of the percentage of natural sand contained in the mixtures on the number of load cycles to develop mm of rutting was analyzed independently. The addition of % natural sand had no statistically significant impact on the number of load cycles to reach mm rut depth for the granite aggregate. However, granite mixtures containing 0% natural sand were statistically different from mixtures containing either no natural sand or % natural sand. Similarly, statistical analyses of limestone mixtures revealed that mixtures containing no natural sand and % natural sand were not statistically different from each other, but both were different from mixtures containing 0% natural sand in terms of number of cycles to mm rut depth. The addition of field sand had no statistically verifiable impact on load cycles to mm rutting for chert gravel mixtures, regardless of the percentage of natural sand in the mix. The effect of neither maximum aggregate size nor relative gradation was statistically significant. Selection of Interim Mixture Evaluation Criteria (Threshold Values) In order to establish interim mixture evaluation criteria for the APA, the number of load cycles to reach different rut depths was considered. Rut depths of mm, mm, mm, and mm were selected as possible failure thresholds. The number of load cycles required to reach these levels could be easily determined since the APA recorded the average rut depth after each load cycle. Figure and show the number of load cycles to achieve these threshold average rut depths. In some cases, mixtures did not reach an average rut depth of or mm before the test was terminated. A value of 000 cycles was denoted as the failure point for these mixtures, even though many more load cycles might have actually been required to obtain these rut depths. The data were separated into two figures identified by the percentage of natural sand in the mixtures, to allow for easier interpretation by reducing the total number of data points in one figure. Only data for mixtures containing no natural sand and % natural sand are presented. The mixtures containing 0% natural sand all failed rapidly in the APA. All but one of these mixtures reached mm average rut depth before,000 cycles. TRB 0 Annual Meeting

14 Rushing, Little, and Garg / CLS / FLS / FLS / CLS / FGN / FGN / FGN / CGN / FGN Number of Load Cycles / CGN / CGN / FGV / CGV / CGN / FLS / CLS / FLS / CLS / GV / FGV 00 / CGV / GV 0 Rut Depth (mm) FIGURE Number of load cycles versus target rut depths for mixtures with no natural sand / CLS / FGN Number of Load Cycles / FLS / CLS / FGN / FLS / CGN / CGV / FGV / CGN / CGN / FGN / CGN / FLS / CLS / FLS / CLS 00 / FGN / GV 0 / GV / FGV / CGV Rut Depth (mm) FIGURE Number of load cycles versus target rut depths for mixtures with % natural sand TRB 0 Annual Meeting

15 Rushing, Little, and Garg 0 0 The data in Figures and indicate a relatively linear rate of rut depth accumulation. Primary rutting (rapid accumulation) generally occurred until a rutting magnitude of about mm. The tertiary phase of rutting was not reached before testing ended. Mixtures containing 0% natural sand rutted faster than mixtures containing either % or no field sand in all cases To provide perspective on these data, researchers compared these results to criteria for agencies that use the APA to accept HMA mixtures. These criteria were obtained from Williams et al. (), and are current as of 00. Other criteria, later than 00, may exist but were not found in the literature search. Results from the APA test with a 0 psi hose pressure are expected to be significantly different from results used to develop criteria for highway pavements. The majority of agencies require APA test specimens to be fabricated with a target air void content of %. Of the twenty-one states reported in Williams paper, only Alabama, Arkansas, and New Jersey test specimens a different air voids content, % air voids, which was the design air void content. Although higher air voids are more indicative of field-placed mixtures, much of the rutting at this air void content may be related to densification and not shear (). Additionally, using the design air void content allows for performance testing of specimens produced from the mix design. Several agencies vary the APA requirements based on the number of design equivalent single axle loads (ESALs). Others have a singular criterion or were listed as still evaluating the APA. Only five states included the APA in the specifications; others used the APA for comparative purposes. All agencies require the application of,000 load cycles during testing. The average maximum allowable rut depth specified is mm. The lowest maximum rut depth is mm (Arkansas and Delaware), while the highest maximum rut depth is mm (Mississippi). Comparison of results from this study with the agency criteria indicated that the high tire pressure APA is much more damaging than the traditionally used APA. After application of,000 load cycles with 0 psi, no mixtures rutted less than mm. Only four mixtures rutted less than mm, while only nine mixtures rutted less than mm. Aside from the mixtures containing 0% natural sand, all mixtures met or exceeded all FAA criteria for HMA mix design. Since the APA is an empirical test that mimics loading, the most valid way to establish threshold values are through correlations with field studies. For example, the criterion recommended by the National Center for Asphalt Technology (NCAT) was less than. mm after,000 cycles at the location high-temperature PG grade (). This criterion was developed through correlation of field data at the NCAT test track. The traffic used in the field test was,000,000 ESALs. These types of correlations indicate the large magnitude of damage each APA cycle induces compared to actual traffic. Ultimately, the rutting threshold established for the heavy wheel load (high tire pressure) APA must be correlated to field results. However, in the interim it is important to establish a viable and realistic interim threshold acceptance rutting relationship between number of passes of the APA and rut depth. In order to do this, we considered the following factors: () the number of applications of aircraft loading relative to highway truck loading, () the tire pressure of airfield traffic relative to highway truck traffic, and () established TRB 0 Annual Meeting

16 Rushing, Little, and Garg 0 correlations of the type summarized in this discussion. Based on these considerations we also considered the results of the mixtures that were tested and the historical performance of these mixtures in airfield situations. The following paragraphs summarize the reasons for establishing the interim criteria. Either or mm of rutting could be used as a criterion for maximum allowable APA rut depth when tested at 0 psi for airport pavements. The number of load cycles at which this level is reached has to be selected to correspond with mixture performance. Using the higher value will still produce a test that can be run in a reasonable timeframe (around hr) and has the ability to differentiate among mixtures. To eliminate the mixtures containing 0% natural sand in this study, the criterion would need to be a maximum of mm rut depth after,000 load cycles. However, this value is still inclusive of some mixtures. Although field performance is unknown, the mixtures containing chert gravel rutted significantly more quickly than mixtures containing quarried aggregate. A criterion of less than mm APA rut depth after,000 cycles would eliminate all but one of the chert gravel mixtures. This criterion would also eliminate one granite mixture containing % natural sand. Based on the data from this study, a reasonable criterion for airport HMA designed for high tire pressure aircraft is less than mm APA rut depth after,000 cycles when tested using 0 psi hose pressure. This criterion would eliminate of the mixtures in this study. However, of the mixtures that failed were not acceptable mixtures because they contained excessive natural sand. Of the other seven failed mixtures, five were chert gravel mixtures that may not be commonly used in airport HMA. Mixture improvements can be made by adjusting the gradation or binder content, and a key advantage of the APA is that the test can be performed in slightly more than one hour and would be easily implementable for HMA mix design. CONCLUSIONS There is a pressing need for a performance test to accompany HMA mix design for airport pavements. Although the Marshall design method uses an empirical index test, a new design method using the SGC relies only on volumetric properties of the compacted mixture for acceptance. This study investigated the suitability of the APA as a test for characterizing HMA for airport pavements subjected to high tire pressure aircraft. From this study we conclude the following: 0 The most significant factor influencing APA rutting is excessive natural sand (0%). Mixtures containing excessive natural sand achieved mm rut depth in less than,000 load cycles. Aggregate type influences APA test results as indicated by the ANOVA. Chert gravel mixtures rutted significantly more quickly than mixtures containing granite or limestone aggregate. The mixtures containing limestone aggregate performed the best of those tested. Chert gravel aggregate had the lowest texture indices according to AIMS testing. Aggregate texture is a better indicator than angularity of rutting resistance according to the Pearson s product moment analysis of APA test results. TRB 0 Annual Meeting

17 Rushing, Little, and Garg The APA test results were not significantly influenced by maximum aggregate size for the mixtures used in this study according to the ANOVA. The APA test results were not significantly influenced by the relative gradation of the mixtures for those used in this study according to the ANOVA. The APA test using a hose pressure of 0 psi rapidly damages HMA specimens. Most mixtures reached the terminal rut depth of mm before,000 cycles were applied. None of the mixtures tested in this study had less than mm rut depth after,000 cycles in the APA. The number of load cycles during the test will likely have to be reduced to produce criterion that will not eliminate a majority of mixtures RECOMMENDATIONS Based on the results of this study, a preliminary criterion of less than mm rut depth after,000 APA load cycles using 0 psi tire pressure is recommended for accepting HMA for high tire pressure aircraft during mix design. This recommendation is limited to the materials used in this study. Further testing should include more aggregate types representative of all regions of the United States. Additional binder types, specifically modified binders, should also be included in future studies. Further, mixtures with proven successful field performance should be evaluated with this criterion, along with mixtures that have shown to rut easily. Further investigations should also determine the correlation of the APA test results with actual in-service pavements. Finally, the repeatability of the APA at high pressures should be investigated to develop tolerances for agency specifications. ACKNOWLEDGEMENTS The authors would like to thank the FAA William J. Hughes Technical Center for sponsoring this work and Mr. Jeff Stein and Mr. Matt Wilson of the FAA for conducting APA testing. The authors would also like to thank Dr. Eyad Masad, Texas A&M University, for his contributions on AIMS testing. Permission was granted to publish this paper by the Director, Geotechnical and Structures Laboratory, ERDC. REFERENCES. Item P-0 Plant Mix Bituminous Pavements (Superpave), Engineering Brief No. A, U.S. Department of Transportation, Federal Aviation Administration, 00.. Zhang, J., Brown, E.R., Kandhal, P.S., and West, R., An Overview of Fundamental and Simulative Performance Tests for Hot Mix Asphalt, Journal of ASTM International, May 00, Vol., No., pp. -.. Williams, R.C., Hill, D.W., and Rottermond, M.P., Utilization of an Asphalt Pavement Analyzer for Hot Mix Asphalt Laboratory Mix Design, Journal of ASTM International, May 00, Vol., No., pp Williams, R.C. and Prowell, B.D., Comparison of Laboratory Wheel-Tracking Test Results with WesTrack Performance, Transportation Research Record, Transportation Research Board, National Research Council, November, pp. -.. West, R.C., Page, G.C., and Murphy, K.H., Evaluation of the Loaded Wheel Tester, Research Report FL/DOT/SMO/-, Florida Department of Transportation, December.. Lai, J.S., Evaluation of Rutting Characteristics of Asphalt Mixes Using the Loaded Wheel Tester, Project No. 0, Georgia Department of Transportation, December. TRB 0 Annual Meeting

18 Rushing, Little, and Garg 0 0. Choubane, B., Page, G.C., and Musselman, J.A., Suitability of Asphalt Pavement Analyzer for Predicting Pavement Rutting, Transportation Research Record, Transportation Research Board, National Research Council, 000, pp. -.. Miller, T., Ksaibati, K., and Farrar, M., Utilizing the Georgia Loaded Wheel Tester to Predict Rutting, Presented at the th Annual meeting of the Transportation Research Board, Washington, D.C., January -,.. WesTrack Forensic Team, Performance of Coarse-Graded Mixes at WesTrack Premature Rutting, Final Report, June.. Rushing, J.F., Development of Criteria for Using the Superpave Gyratory Compactor to Design Airport Pavement Mixtures, Master s Thesis, Mississippi State University, Mississippi State, MS, 00.. Standards for Specifying Construction of Airports, Advisory Circular 0/0 E, U.S. Department of Transportation, Federal Aviation Administration, 00.. Kandhal, P., Motter, J., and Khatri, M., Evaluation of Particle Shape and Texture: Manufactured Versus Natural Sands, NCAT -0, National Center for Asphalt Technology, Auburn University, Auburn, AL,.. Masad, E., Al-Rousan, T., Button, J., Little, D., and Tutumluer, E. (00). Test Methods for Characterizing Aggregate Shape, Texture and Angularity, NCHRP -0A Final Report, Report Number, National Cooperative Highway Research Program, National Research Council, Washington, D.C.. Asphalt Institute, Mix Design Methods for Asphalt, th ed., MS-0, Asphalt Institute. Lexington, KY,.. Button, J.W, Perdomo, D., and Lytton, R.L., Influence of Aggregate on Rutting in Asphalt Concrete Pavements, Transportation Research Record, TRB, National Research Council, Washington, D.C., 0, pp. -.. Ahlrich, R.C., Influence of Aggregate Gradation and Particle Shape/Texture on Permanent Deformation of Hot Mix Asphalt Pavements, Technical Report GL--, U.S. Army Engineer Waterways Experiment Station, Vicksburg, MS,.. Chou, Y.T., Analysis of Permanent Deformations of Flexible Airport Pavements, WES TR S--, U.S. Army Waterways Experiment Station, Vicksburg, MS,.. Zhang, J., Cooley, Jr., L.A., and Kandhal, P.S., Comparison of Fundamental and Simulative Test Methods for Evaluating Permanent Deformation of Hot Mix Asphalt, Transportation Research Record, TRB, National Research Council, Washington, D.C., 00, pp TRB 0 Annual Meeting